Adaptive graph convolutional imputation network for environmental sensor data recovery
Environmental sensors are essential for tracking weather conditions and changing trends, thus preventing adverse effects on species and environment. Missing values are inevitable in sensor recordings due to equipment malfunctions and measurement errors. Recent representation learning methods attempt...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2022-11-01
|
| Series: | Frontiers in Environmental Science |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fenvs.2022.1025268/full |
| _version_ | 1798018055246184448 |
|---|---|
| author | Fanglan Chen Dongjie Wang Shuo Lei Jianfeng He Yanjie Fu Chang-Tien Lu |
| author_facet | Fanglan Chen Dongjie Wang Shuo Lei Jianfeng He Yanjie Fu Chang-Tien Lu |
| author_sort | Fanglan Chen |
| collection | DOAJ |
| description | Environmental sensors are essential for tracking weather conditions and changing trends, thus preventing adverse effects on species and environment. Missing values are inevitable in sensor recordings due to equipment malfunctions and measurement errors. Recent representation learning methods attempt to reconstruct missing values by capturing the temporal dependencies of sensor signals as handling time series data. However, existing approaches fall short of simultaneously capturing spatio-temporal dependencies in the network and fail to explicitly model sensor relations in a data-driven manner. In this work, we propose a novel Adaptive Graph Convolutional Imputation Network for missing value imputation in environmental sensor networks. A bidirectional graph convolutional gated recurrent unit module is introduced to extract spatio-temporal features which takes full advantage of the available observations from the target sensor and its neighboring sensors to recover the missing values. In addition, we design an adaptive graph learning layer that learns a sensor network topology in an end-to-end framework, in which no prior network information is needed for capturing spatial dependencies. Extensive experiments on three real-world environmental sensor datasets (solar radiation, air quality, relative humidity) in both in-sample and out-of-sample settings demonstrate the superior performance of the proposed framework for completing missing values in the environmental sensor network, which could potentially support environmental monitoring and assessment. |
| first_indexed | 2024-04-11T16:17:02Z |
| format | Article |
| id | doaj.art-6614c701c82a4ff28ef04b54e9356631 |
| institution | Directory Open Access Journal |
| issn | 2296-665X |
| language | English |
| last_indexed | 2024-04-11T16:17:02Z |
| publishDate | 2022-11-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Environmental Science |
| spelling | doaj.art-6614c701c82a4ff28ef04b54e93566312022-12-22T04:14:28ZengFrontiers Media S.A.Frontiers in Environmental Science2296-665X2022-11-011010.3389/fenvs.2022.10252681025268Adaptive graph convolutional imputation network for environmental sensor data recoveryFanglan Chen0Dongjie Wang1Shuo Lei2Jianfeng He3Yanjie Fu4Chang-Tien Lu5Department of Computer Science, Virginia Tech, Falls Church, VA, United StatesDepartment of Computer Science, University of Central Florida, Orlando, FL, United StatesDepartment of Computer Science, Virginia Tech, Falls Church, VA, United StatesDepartment of Computer Science, Virginia Tech, Falls Church, VA, United StatesDepartment of Computer Science, University of Central Florida, Orlando, FL, United StatesDepartment of Computer Science, Virginia Tech, Falls Church, VA, United StatesEnvironmental sensors are essential for tracking weather conditions and changing trends, thus preventing adverse effects on species and environment. Missing values are inevitable in sensor recordings due to equipment malfunctions and measurement errors. Recent representation learning methods attempt to reconstruct missing values by capturing the temporal dependencies of sensor signals as handling time series data. However, existing approaches fall short of simultaneously capturing spatio-temporal dependencies in the network and fail to explicitly model sensor relations in a data-driven manner. In this work, we propose a novel Adaptive Graph Convolutional Imputation Network for missing value imputation in environmental sensor networks. A bidirectional graph convolutional gated recurrent unit module is introduced to extract spatio-temporal features which takes full advantage of the available observations from the target sensor and its neighboring sensors to recover the missing values. In addition, we design an adaptive graph learning layer that learns a sensor network topology in an end-to-end framework, in which no prior network information is needed for capturing spatial dependencies. Extensive experiments on three real-world environmental sensor datasets (solar radiation, air quality, relative humidity) in both in-sample and out-of-sample settings demonstrate the superior performance of the proposed framework for completing missing values in the environmental sensor network, which could potentially support environmental monitoring and assessment.https://www.frontiersin.org/articles/10.3389/fenvs.2022.1025268/fullmissing data imputationadaptive graph learningspatio-temporal deep learninggraph neural networkenvironmental sensor network |
| spellingShingle | Fanglan Chen Dongjie Wang Shuo Lei Jianfeng He Yanjie Fu Chang-Tien Lu Adaptive graph convolutional imputation network for environmental sensor data recovery Frontiers in Environmental Science missing data imputation adaptive graph learning spatio-temporal deep learning graph neural network environmental sensor network |
| title | Adaptive graph convolutional imputation network for environmental sensor data recovery |
| title_full | Adaptive graph convolutional imputation network for environmental sensor data recovery |
| title_fullStr | Adaptive graph convolutional imputation network for environmental sensor data recovery |
| title_full_unstemmed | Adaptive graph convolutional imputation network for environmental sensor data recovery |
| title_short | Adaptive graph convolutional imputation network for environmental sensor data recovery |
| title_sort | adaptive graph convolutional imputation network for environmental sensor data recovery |
| topic | missing data imputation adaptive graph learning spatio-temporal deep learning graph neural network environmental sensor network |
| url | https://www.frontiersin.org/articles/10.3389/fenvs.2022.1025268/full |
| work_keys_str_mv | AT fanglanchen adaptivegraphconvolutionalimputationnetworkforenvironmentalsensordatarecovery AT dongjiewang adaptivegraphconvolutionalimputationnetworkforenvironmentalsensordatarecovery AT shuolei adaptivegraphconvolutionalimputationnetworkforenvironmentalsensordatarecovery AT jianfenghe adaptivegraphconvolutionalimputationnetworkforenvironmentalsensordatarecovery AT yanjiefu adaptivegraphconvolutionalimputationnetworkforenvironmentalsensordatarecovery AT changtienlu adaptivegraphconvolutionalimputationnetworkforenvironmentalsensordatarecovery |